Clean task data

This document provides the code for the analyses reported in:

Cosme et al. (Preprint) Testing the adolescent social reorientation model using hierarchical growth curve modeling with parcellated fMRI data

This script cleans the self/other task data for the analyses reported in supplementary material.

load packages

library(tidyverse)
library(stringr)

load task data

# define variables and paths
datadir = "/Volumes/psych-cog/dsnlab/SFIC/behavior/task/"
sub_pattern = "L[0-3]{1}([0-9]{2})"
waves = list.files(datadir, pattern = "t[1-3]{1}")

# initialize data frame
data = data.frame()

# load data
for (wave in waves) {
  wavedir = paste0(datadir, wave, "/RTs_txt/")
  subs = list.files(wavedir, pattern = sub_pattern)
  for (sub in subs) {
    filename = paste0(wavedir, sub)
    tmp = tryCatch(read.table(filename, header = FALSE, sep = "\t", fill = TRUE, skipNul = TRUE) %>%
                              mutate(file = filename,
                                     wave = wave,
                                     V1 = str_replace(V1, "\\*", "000")) %>%
                              extract(file, "subjectID", sub_pattern) %>%
                              mutate(subjectID = paste0("s0", subjectID),
                                     subFile = sub) %>%
                              extract(V1, "script", ".*000(.*)_.*", remove = FALSE) %>% 
                              extract(V1, "buttons", "([a-zA-Z0-9= ]+).*000.*", remove = FALSE) %>%
                              mutate(order = ifelse(grepl("vp|vn|pp|pn", V4), .$script[grep("hv", script)], NA),
                                     yn.key = ifelse(grepl("vp|vn|pp|pn", V4), .$buttons[grep("1|2|3|4|g|r|b|y", buttons)], NA)) %>%
                              filter(!is.na(order)) %>%
                              filter(!V1 == "\026\001\020\f\003\f\b8list\001reco\030\001pjstenum\004deft") %>%
                              rename("rt" = V2,
                                     "trial.end" = V3,
                                     "stimulus" = V4,
                                     "response" = V5) %>%
                              mutate(trial = row_number()) %>%
                              mutate_all(as.character) %>%
                              select(subFile, subjectID, wave, order, yn.key, trial, rt, trial.end, stimulus, response),
                   error = function(e) NULL)
      
    data = bind_rows(data, tmp)
  }
}

load age and merge

# load 
age = read.csv("/Volumes/psych-cog/dsnlab/SFIC_data_sheets/SFIC_age.pds.gender.csv") %>%
  rename("subjectID" = SID,
         "wave" = wavenum,
         "gender" = Gender) %>%
  mutate(subjectID = sprintf("s%03d", subjectID),
         wave = paste0("t", wave)) %>%
  select(-c(DOB, session))

# merge
data = data %>%
  left_join(., age)

check number of trials per participant

# check number of trials for all participants
data %>%
  group_by(subjectID, wave) %>%
  summarise(n.trials = n()) %>%
  arrange(desc(n.trials))

# check irregular number of trials
data %>%
  filter(subjectID %in% c("s030", "s035", "s018", "s027", "s063", "s081")) %>%
  group_by(subFile, subjectID, wave) %>%
  summarise(n.trials = n()) %>%
  arrange(desc(n.trials))

select correct responses

• s030 t3 - From the log: “Ran Self two times. The first time the computer was not recording his responses. The second run is good and what should be used.”
  • There are actually 3 runs of the self task in the file
    
  • The order for this participant is hpsphvsv
    
  • The first is order hpsphvsv and lacks responses
    
  • The second is order hvsvhpsp and has responses –> this is a copy of s019 t3 (timing and responses match)
  • The third is order hpsphvsv and has responses –> using this run
• s035 t2 - No note in the log, but L235 -2 has the correct order hpsphvsv –> using L235 -2 (second start; yn key = 12)
• s018 t3 - No note in the log, but L318-928226 is the subject ID listed –> using L318-928226.txt
  
• s027 t1 - No note in the log, but the first run has ! as a response for many of the trials –> using second run
  
• s063 t1 - No note in the log (?), but was restarted –> remove 19 trials (yn key = 34)
• s081 t3 - No note in the log, but the first run only has 16 trials and no responses –> using second run

s030_t3 = data %>%
  filter(subjectID == "s030" & wave == "t3") %>%
  slice(-1:-160) %>%
  mutate(order = "hpsphvsv") %>%
  mutate(trial = as.character(row_number()))

s035_t2 = data %>%
  filter(subFile == "L235 - 2.txt") %>%
  slice(-1:-16) %>%
  mutate(trial = as.character(row_number()),
         yn.key = "12")

s027_t1 = data %>%
  filter(subjectID == "s027" & wave == "t1") %>%
  slice(-1:-80) %>%
  mutate(trial = as.character(row_number()),
         yn.key = "34")

s063_t1 = data %>%
  filter(subjectID == "s063" & wave == "t1") %>%
  slice(-1:-19) %>%
  mutate(trial = as.character(row_number()),
         yn.key = "34")

s081_t3 = data %>%
  filter(subjectID == "s081" & wave == "t3") %>%
  slice(-1:-16) %>%
  mutate(trial = as.character(row_number()))

# filter out subjects original data and then bind subsetted data
data1 = data %>%
  filter(!subFile == "L318-085822.txt") %>%
  filter(!(subjectID  == "s030" & wave == "t3")) %>%
  filter(!(subjectID  == "s035" & wave == "t2")) %>%
  filter(!(subjectID  == "s027" & wave == "t1")) %>%
  filter(!(subjectID  == "s063" & wave == "t1")) %>%
  filter(!(subjectID  == "s081" & wave == "t3")) %>%
  bind_rows(., s030_t3) %>%
  bind_rows(., s035_t2) %>%
  bind_rows(., s027_t1) %>%
  bind_rows(., s063_t1) %>%
  bind_rows(., s081_t3)

# check number of rows
nrow(data1) == nrow(data) - 80 - 160 - 96 - 80 - 19 - 16

## [1] TRUE

check button reponses

# unique button responses for all participants
(buttons = data1 %>% 
   group_by(wave, yn.key, subjectID, response) %>% 
   summarize(n = n()) %>%
   arrange(wave, subjectID))

# unique responses across participants
unique(buttons$response)

##  [1] ""               "g"              "gg"             "r"             
##  [5] "rg"             "rr"             "ggg"            "gr"            
##  [9] "ggr"            "rgg"            "rgr"            "rrr"           
## [13] "3"              "33"             "4"              "44"            
## [17] "34"             "43"             "1"              "12"            
## [21] "2"              "21"             "22"             "11"            
## [25] "3333"           "33333"          "33333333"       "333333333"     
## [29] "333333333333"   "33333333333333" "43333333333"    "b"             
## [33] "by"             "y"              "yy"             "yyyyyyyyyyyy"  
## [37] "112"

manual changes

Manual changes to some text files were made to facilitate loading the data into r. When the yes/no key was on a separate line from the condition order, the line was removed. When strange characters were included before the yes/no key they were removed.

For all other errors, these will be addressed directly in r. The original, raw data is still intact as binary files in SFIC/behavor/[wave]/RTs.

T1 changes

• s027 - removed line, removed character in line 1
  
• s070 - removed character
  
• s019 - removed line
  
• s039 - removed line
  
• s041 - removed line and character
• s046 - removed line
  
• s063 - removed line and character
• s067 - removed line
  
• s079 - removed line

T2 changes

• s001 - removed line
• s011 - removed character
• s017 - removed character
• s078 - copied from line 1
  
• s057 - removed character
  
• s070 - removed line
• s073 - removed line

T3 changes

• s019 - removed line
• s035 - removed line
• s057 - removed character
• s064 - entered first line to a new line
• s090 - removed character

recode yes/no key

Relevant notes from T1 log files

• s001 - no responses recorded
• s002 - r = yes, g = no
• s005 - g = yes, r = no
• s006 - g = yes, r = no
• s008 - g = yes, r = no
• s009 - r = yes, g = no
• s027 - re-ran scan to collect responses
• s034 - 3 = yes, 4 = no (difficulty hearing phrases)
• s093 - missing last 19s during rest block

Relevant notes from T2 log files

• s045 - Restarted self run after 3 mins because “no” button was not working
• s070 - Do Not Use. Subject scooted WAY down by this scan. Registration from HiRes is no good

Relevant notes from T3 log files

• s030 - Ran Self two times. The first time the computer was not recording his responses. The second run is good and what should be used.
  • Note this is actually the third run (second run is a copy of s019 t3)

# check keys
unique(data1$yn.key)

##  [1] "grrg"                                                                                  
##  [2] "gr"                                                                                    
##  [3] "rg"                                                                                    
##  [4] "grrggggggggggggg"                                                                      
##  [5] NA                                                                                      
##  [6] "43"                                                                                    
##  [7] "34"                                                                                    
##  [8] "grrggr"                                                                                
##  [9] "grr"                                                                                   
## [10] "3333333334"                                                                            
## [11] "3434"                                                                                  
## [12] "4=no 3=yes"                                                                            
## [13] "34343434"                                                                              
## [14] "344"                                                                                   
## [15] "12"                                                                                    
## [16] "1212121212121212112222211121212121212121212121212111112"                               
## [17] "yyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyy"
## [18] "3=yes 4=no"                                                                            
## [19] "by"                                                                                    
## [20] "122"                                                                                   
## [21] "b12"                                                                                   
## [22] "121212"                                                                                
## [23] "121"                                                                                   
## [24] "2"                                                                                     
## [25] "1212"                                                                                  
## [26] "111222"                                                                                
## [27] "bb12"                                                                                  
## [28] "b"

# select responses to recode
yn.gr = c("grrggggggggggggg", "grrggr", "grr")
yn.34 = c("3333333334", "3434", "4=no 3=yes", "34343434", "344", "3=yes 4=no")
yn.12 = c("1212121212121212112222211121212121212121212121212111112", "yyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyy", "122", "b12", "121212", "121", "2", "1212", "111222", "bb12", "b")

# recode responses
data2 = data1 %>%
  mutate(yn.key = ifelse(yn.key %in% yn.gr, "gr",
                  ifelse(yn.key %in% yn.34, "34",
                  ifelse(yn.key %in% yn.12, "12", yn.key))),
         yn.key = ifelse(wave == "t1" & subjectID %in% c("s002"), "rg",
                  ifelse(wave == "t2" & subjectID == "s065", "12", yn.key)))

# check keys
unique(data2$yn.key)

## [1] "rg" "gr" NA   "43" "34" "12" "by"

# summarize response types
data2 %>%
  group_by(yn.key) %>%
  summarize(n = n())

# summarize response types by wave
data2 %>%
  group_by(yn.key, wave) %>%
  summarize(n = n()) %>%
  arrange(wave) %>%
  select(wave, everything())

# summarize response types by subject and wave
data2 %>%
  group_by(subjectID, yn.key, wave) %>%
  summarize(n = n()) %>%
  arrange(wave) %>%
  select(wave, everything())

check responses for missing button orders

• based on the breakdown by wave, it looks like the y/n order for T3 is very likely 12
• for the one T1 participant missing the y/n order, there’s a higher probability that it was gr, but we can’t be sure. How do we want to deal with this data?

buttons %>% 
  filter(is.na(yn.key)) %>%
  arrange(wave, subjectID)

missing.buttons.t1 = buttons %>% 
  filter(is.na(yn.key) & wave == "t1")

missing.buttons.t2 = buttons %>% 
  filter(is.na(yn.key) & wave == "t2")

missing.buttons.t3 = buttons %>% 
  filter(is.na(yn.key) & wave == "t3")

missing.buttons.t1 = unique(missing.buttons.t1$subjectID)
missing.buttons.t2 = unique(missing.buttons.t2$subjectID)
missing.buttons.t3 = unique(missing.buttons.t3$subjectID)

describe rt

• exclude 0 because this is only used for missing responses

data2.g0 = data2 %>%
  mutate(rt = as.numeric(rt)) %>%
  select(subjectID, wave, rt, response)

data2.g0 %>%
  select(rt) %>%
  filter(!rt == 0) %>%
  psych::describe(quant = c(.25, .75))

# plot RTs for all responses
data2.g0 %>%
  mutate(`previous missed` = ifelse(lag(rt) == 0, "yes", "no")) %>%
  filter(!is.na(`previous missed`) & !rt == 0) %>%
  ggplot(aes("", rt, color = `previous missed`)) +
    geom_boxplot() + 
    geom_jitter(width = 0.1, alpha = .1) + 
    facet_grid(~wave) +
    labs(x = "") + 
    theme_minimal()

# plot RTs by response
data2.g0 %>%
  mutate(`previous missed` = ifelse(lag(rt) == 0, "yes", "no"),
         n.char = nchar(response)) %>%
  filter(!is.na(`previous missed`) & !rt == 0) %>%
  ggplot(aes("", rt, color = `previous missed`)) +
    geom_boxplot() + 
    geom_jitter(width = 0.1, alpha = .1) + 
    facet_grid(n.char ~wave) +
    labs(x = "") + 
    theme_minimal()

# plot RTs for the first time hearing each stimulus only
data %>%
  mutate(rt = as.numeric(rt),
         `previous missed` = ifelse(lag(rt) == 0, "yes", "no"),
         n.char = nchar(response),
         trial = as.numeric(trial),
         keep = ifelse(trial %in% c(1:20, 41:60), 1, 0)) %>%
  filter(!is.na(`previous missed`) & !rt == 0 & keep == 1) %>%
  ggplot(aes("", rt, color = `previous missed`)) +
    geom_boxplot() + 
    geom_jitter(width = 0.1, alpha = .1) + 
    facet_grid(~wave) +
    labs(x = "") + 
    theme_minimal()

# density plots
data %>%
  mutate(rt = as.numeric(rt),
         `previous missed` = ifelse(lag(rt) == 0, "yes", "no"),
         n.char = nchar(response),
         trial = as.numeric(trial),
         keep = ifelse(trial %in% c(1:20, 41:60), 1, 0)) %>%
  filter(!is.na(`previous missed`) & !rt == 0 & keep == 1) %>%
  ggplot(aes(rt, fill = `previous missed`)) +
    geom_density(alpha = .75) + 
    facet_grid(~wave) +
    labs(x = "") + 
    theme_minimal()

# plot RTs of trials where the previous trial was missed and there are 2 responses
data2.g0 %>%
  group_by(subjectID, wave) %>%
  mutate(n.char = nchar(response)) %>%
  filter(lag(rt) == 0 & n.char > 1) %>%
  ggplot(aes("", rt)) +
  geom_violin() + 
  geom_jitter(width = 0.1, alpha = .25) + 
  facet_grid(~wave) + 
  labs(x = "") + 
  theme_minimal()

# plot RTs of trials where the previous trial was missed is one response
data2.g0 %>%
  group_by(subjectID, wave) %>%
  mutate(n.char = nchar(response)) %>%
  filter(lag(rt) == 0 & n.char == 1) %>%
  ggplot(aes("", rt)) +
  geom_violin() + 
  geom_jitter(width = 0.1, alpha = .25) + 
  facet_grid(~wave) + 
  labs(x = "") + 
  theme_minimal()

quantile(filter(data2.g0, wave == "t1")$rt, seq(0,1,.1))

##     0%    10%    20%    30%    40%    50%    60%    70%    80%    90%   100% 
## 0.0000 0.0000 0.6774 1.7660 1.9300 2.0590 2.1840 2.3110 2.4602 2.6491 2.9740

quantile(filter(data2.g0, wave == "t2")$rt, seq(0,1,.1))

##     0%    10%    20%    30%    40%    50%    60%    70%    80%    90%   100% 
## 0.0000 0.0000 1.4990 1.7757 1.9016 2.0280 2.1494 2.2610 2.4062 2.6161 2.9790

quantile(filter(data2.g0, wave == "t3")$rt, seq(0,1,.1))

##     0%    10%    20%    30%    40%    50%    60%    70%    80%    90%   100% 
## 0.0000 0.1270 1.6150 1.7787 1.8940 1.9890 2.0810 2.1970 2.3372 2.5090 2.9670

fix responses

There are missing responses and multiple responses.

Coding rules:

• if the response for trial 1 is missing and trial 2 is < 1s and has multiple response, code trial 1 as the first response and trial 2 as the second (or third) response
• if the response for trial 1 is missing and the trial 2 RT < .75s, code trial 1 as the first response
  • if trial 2 only has one response, trial 2 will no longer have a response
    
  • if trial 2 has more than one response, trial 2 will be the second
• if multiple responses, use the last response

(data3 = data2 %>%
  mutate(rt = as.numeric(rt),
         trial = as.numeric(trial),
         quick = ifelse(rt <= 1 & rt > 0, 1, 0),
         n.char = nchar(response),
         first = substring(response, 1, 1),
         second = substring(response, 2, 2),
         third = substring(response, 3, 3)))

# describe rts
data3 %>%
  mutate(rt.bin = ifelse(rt > 0 & rt <= .25, "lt.25",
                  ifelse(rt > .25 & rt <= .5, "lt.5",
                  ifelse(rt > .5 & rt <= .75, "lt.75",
                  ifelse(rt > .75 & rt <= 1, "lt.1", "gt.1"))))) %>%
  group_by(rt.bin, wave) %>%
  summarize(n = n()) %>%
  spread(wave, n)

# describe average RT by subject
data3 %>%
  filter(rt > 0) %>%
  group_by(subjectID, wave) %>%
  summarize(q10 = quantile(rt, .1, na.rm = TRUE),
            q25 = quantile(rt, .25, na.rm = TRUE),
            q50 = quantile(rt, .5, na.rm = TRUE),
            mean.rt = mean(rt, na.rm = TRUE),
            threshold = mean.rt - sd(rt, na.rm = TRUE))

# summarize issues
data3 %>%
  group_by(subFile) %>%
  mutate(issue = ifelse(response == "" & lead(rt) < .5 & lead(n.char) == 1, "missed, RT <.5, 1 response",
                 ifelse(response == "" & lead(rt) >= .5 & lead(rt) < .75 & lead(n.char) == 1, "missed, .5< RT <.75, 1 response",
                 ifelse(response == "" & lead(rt) >= .75 & lead(rt) < 1 & lead(n.char) == 1, "missed, .75< RT <1, 1 response",
                 ifelse(response == "" & lead(rt) < 1 & lead(n.char) == 2, "missed, RT <1, 2 responses",
                 ifelse(response == "" & lead(rt) < 1 & lead(n.char) == 3, "missed, RT <1, 3 responses",
                 ifelse(lead(rt) < 1 & lead(n.char) == 1, "not missed, RT <1, 1 response", 
                 ifelse(lead(rt) < 1 & lead(n.char) > 1, "not missed, RT <1, >1 response", NA)))))))) %>%
  group_by(issue, wave) %>%
  summarize(n = n()) %>%
  spread(wave, n)

# check potential recoding responses
data.check = data3 %>%
  group_by(subFile) %>%
  mutate(response.fixed = ifelse(lag(response) == "" & rt <= .75 & n.char == 1, "", response), #missed, RT <.75, 1 response
         response.fixed = ifelse(response == "" & lead(rt) <= .75 & lead(n.char) == 1, lead(first), #missed, RT <.75, 1 response
                          ifelse(response == "" & lead(rt) > .75 & lead(rt) < 1 & lead(n.char) == 1, response, #missed, .75< RT <1, 1 response
                          ifelse(response == "" & lead(rt) < 1 & lead(n.char) == 2, lead(first), #missed, RT <1, 2 responses
                          ifelse(response == "" & lead(rt) < 1 & lead(n.char) == 3, lead(first), #missed, RT <1, 3 responses
                          ifelse(n.char == 2, second, #2 responses
                          ifelse(n.char == 3, third, #3 responses
                          ifelse(trial == 80, response, 9999))))))), #response
         response.fixed = ifelse(response == "" & lead(rt) > .75 & lead(rt) < 1 & lead(n.char) == 1, "00", response.fixed), #missed, .75< RT <1, 1 response
         response.check = ifelse(response.fixed == 9999, 0, 1), 
         response.check = ifelse(lag(response.check) == 1, 1, response.check)) %>%
  filter(response.check == 1)

# recode responses
data4 = data3 %>%
  group_by(subFile) %>%
  mutate(response.fixed = ifelse(lag(response) == "" & rt <= .75 & n.char == 1, "", response), #missed, RT <.75, 1 response
         response.fixed = ifelse(response == "" & lead(rt) <= .75 & lead(n.char) == 1, lead(first), #missed, RT <.75, 1 response
                          ifelse(response == "" & lead(rt) > .75 & lead(rt) < 1 & lead(n.char) == 1, response, #missed, .75< RT <1, 1 response
                          ifelse(response == "" & lead(rt) < 1 & lead(n.char) == 2, lead(first), #missed, RT <1, 2 responses
                          ifelse(response == "" & lead(rt) < 1 & lead(n.char) == 3, lead(first), #missed, RT <1, 3 responses
                          ifelse(n.char == 2, second, #2 responses
                          ifelse(n.char >= 3, third, #3 responses
                          ifelse(trial == 80, response, response.fixed))))))))

# summarize affected trials
data3 %>%
  group_by(subFile) %>%
  mutate(issue = ifelse(response == "" & lead(rt) <= .75 & lead(n.char) == 1, "issue",
                 ifelse(response == "" & lead(rt) > .75 & lead(rt) < 1 & lead(n.char) == 1, "issue",
                 ifelse(response == "" & lead(rt) < 1 & lead(n.char) == 2, "issue",
                 ifelse(response == "" & lead(rt) < 1 & lead(n.char) == 3, "issue", "no issue"))))) %>%
  filter(!is.na(issue)) %>% #filter out missing last trials
  group_by(issue, wave) %>%
  summarize(n = n()) %>%
  group_by(wave) %>%
  mutate(percent = round((n / sum(n)) * 100, 1)) %>%
  group_by(issue) %>%
  mutate(n_issue = sum(n)) %>%
  ungroup() %>%
  mutate(percent_issue = round((664 / (664 + 11959)) * 100, 1))

# print unique responses
unique(data4$response.fixed)

##  [1] ""  "g" "r" NA  "3" "4" "1" "2" "y" "b"

# get percentage of missed trials
(missed = data4 %>%
  mutate(response.fixed = ifelse(response.fixed %in% c("g", "r", "3", "4", "1", "2", "b", "y"), "response", NA)) %>%
  group_by(subjectID, wave, response.fixed) %>%
  summarize(n = n()) %>%
  spread(response.fixed, n) %>%
  mutate(percent = round((`<NA>` / 80) * 100, 1)) %>%
  arrange(desc(percent)))

missed %>%
  ggplot(aes(percent)) +
    geom_histogram() + 
    theme_minimal()

recode responses to yes/no

• if s017 is coded “rg” social and academic status biases become negative
• if missing t1 data are coded “21” status biases become < 50%
• if missing t3 data are coded “21” status biases become < 50%

unique(data4$yn.key)

## [1] "rg" "gr" NA   "43" "34" "12" "by"

missing.yn = data4 %>%
  filter(is.na(yn.key))

data5 = data4 %>%
  mutate(yn.key = ifelse(is.na(yn.key) & wave == "t1" & subjectID == "s017" , "gr",
                  ifelse(is.na(yn.key) & wave == "t2" & subjectID == "s018" , "12",
                  ifelse(is.na(yn.key) & wave == "t2" & subjectID == "s042" , "12",
                  ifelse(is.na(yn.key) & wave == "t3", "12", yn.key)))),
         response.yn = ifelse(yn.key == "gr" & response.fixed == "g", "yes",
                       ifelse(yn.key == "gr" & response.fixed == "r", "no",
                       ifelse(yn.key == "rg" & response.fixed == "r", "yes",
                       ifelse(yn.key == "rg" & response.fixed == "g", "no",
                       ifelse(yn.key == "43" & response.fixed == "4", "yes",
                       ifelse(yn.key == "43" & response.fixed == "3", "no",
                       ifelse(yn.key == "34" & response.fixed == "3", "yes",
                       ifelse(yn.key == "34" & response.fixed == "4", "no",
                       ifelse(yn.key == "12" & response.fixed == "1", "yes",
                       ifelse(yn.key == "12" & response.fixed == "2", "no", NA)))))))))))

unique(data5$response.yn)

## [1] NA    "no"  "yes"

recode stimuli

unique(data5$order)

## [1] "hvsvhpsp" "svhvsphp" "hpsphvsv" "sphpsvhv"

data6 = data5 %>%
  mutate(trial = as.numeric(trial),
         target = ifelse(grepl("^h", order) & trial %in% c(1:20, 41:60), "other",
                  ifelse(grepl("^h", order) & trial %in% c(21:40, 61:80), "self",
                  ifelse(grepl("^s", order) & trial %in% c(1:20, 41:60), "self",
                  ifelse(grepl("^s", order) & trial %in% c(21:40, 61:80), "other", NA)))),
         domain = ifelse(grepl("^v", stimulus), "academic", "social"),
         valence = ifelse(grepl("vp|pp", stimulus), "positive", "negative"))

unique(data6$target)

## [1] "other" "self"

unique(data6$domain)

## [1] "academic" "social"

unique(data6$valence)

## [1] "positive" "negative"

create subjective social status/verbal academic competence metric

• Note that s029 t3 only has responses for the other academic block; no responses were recoded for the other blocks

data.status = data6 %>%
  group_by(subjectID, wave, domain, target, valence, response.yn) %>%
  summarize(age = mean(age),
            n = n()) %>%
  spread(response.yn, n) %>%
  mutate(no = ifelse(is.na(no), 0, no),
         yes = ifelse(is.na(yes), 0, yes),
         n = no + yes,
         status = ifelse(valence == "negative", no,
                  ifelse(valence == "positive", yes, NA))) %>%
  group_by(subjectID, wave, domain, target) %>%
  mutate(n_response = sum(n, na.rm = TRUE),
         n_NA = sum(`<NA>`, na.rm = TRUE),
         task_net = sum(status, na.rm = TRUE),
         task_percent = (task_net / n_response) * 100,
         task_percent1 = ifelse(is.nan(task_percent), NA, task_percent)) %>%
  select(-c(valence, yes, no, n, status, `<NA>`))

data.status %>%
  ggplot(aes(wave, task_percent, color = target)) +
    geom_violin() + 
    geom_jitter(width = 0.2, alpha = .2) +
    facet_grid(target ~ domain) +
    theme_minimal() +
    theme(legend.position = "none")

check status bias

# social and academic
data.status %>%
  filter(target == "self") %>%
  arrange(task_percent)

# social only
data.status %>%
  filter(target == "self" & domain == "social") %>%
  arrange(task_percent)

# for participants missing yn.key
data.status %>%
  filter(target == "self" & ((subjectID %in% missing.buttons.t1 & wave == "t1") | (subjectID %in% missing.buttons.t2 & wave == "t2") | (subjectID %in% missing.buttons.t3 & wave == "t3"))) %>%
  arrange(task_percent)

data.status %>%
  filter(target == "self" & domain == "social" & ((subjectID %in% missing.buttons.t1 & wave == "t1") | (subjectID %in% missing.buttons.t2 & wave == "t2") | (subjectID %in% missing.buttons.t3 & wave == "t3")))

data.status %>%
  filter(target == "self" & ((subjectID %in% missing.buttons.t1 & wave == "t1") | (subjectID %in% missing.buttons.t2 & wave == "t2") | (subjectID %in% missing.buttons.t3 & wave == "t3"))) %>%
  arrange(subjectID)

save

data.status %>%
  select(subjectID, wave, domain, target, task_percent) %>%
  unique() %>%
  saveRDS(., "../../data/task.RDS")